任务的划分
DAGScheduler类的handleJobSubmitted方法中,有一个提交阶段的的方法:
var finalStage: ResultStage = null
……
finalStage = createResultStage(finalRDD, func, partitions, jobId, callSite)
……
submitStage(finalStage)
submitStage方法用于提交最终的ResultStage阶段,由于在最终的ResultStage可能包含了多个上级阶段,所以此处就相当于是提交整个应用程序的全部阶段。查看一下该方法的源码:
private def submitStage(stage: Stage): Unit = {
val jobId = activeJobForStage(stage)
if (jobId.isDefined) {
logDebug(s"submitStage($stage (name=${stage.name};" +
s"jobs=${stage.jobIds.toSeq.sorted.mkString(",")}))")
if (!waitingStages(stage) && !runningStages(stage) && !failedStages(stage)) {
val missing = getMissingParentStages(stage).sortBy(_.id)
logDebug("missing: " + missing)
if (missing.isEmpty) {
logInfo("Submitting " + stage + " (" + stage.rdd + "), which has no missing parents")
submitMissingTasks(stage, jobId.get)
} else {
for (parent <- missing) {
submitStage(parent)
}
waitingStages += stage
}
}
} else {
abortStage(stage, "No active job for stage " + stage.id, None)
}
}
该方法的内部核心逻辑是先获取当前阶段的的所有父级阶段,如果其父级阶段为空那么直接执行submitMissingTasks方法,如果不为空,那么递归执行submitStage方法,只不过传入的参数是当前阶段的父级阶段,一直递归直到找到没有上级阶段的阶段,最终没有上级阶段的那个阶段会执行submitMissingTasks方法。下面查看一下该方法的核心源码部分:
private def submitMissingTasks(stage: Stage, jobId: Int): Unit = {
……
val partitionsToCompute: Seq[Int] = stage.findMissingPartitions()
……
val tasks: Seq[Task[_]] = try {
val serializedTaskMetrics = closureSerializer.serialize(stage.latestInfo.taskMetrics).array()
stage match {
case stage: ShuffleMapStage =>
stage.pendingPartitions.clear()
partitionsToCompute.map { id =>
val locs = taskIdToLocations(id)
val part = partitions(id)
stage.pendingPartitions += id
new ShuffleMapTask(stage.id, stage.latestInfo.attemptNumber,
taskBinary, part, locs, properties, serializedTaskMetrics, Option(jobId),
Option(sc.applicationId), sc.applicationAttemptId, stage.rdd.isBarrier())
}
case stage: ResultStage =>
partitionsToCompute.map { id =>
val p: Int = stage.partitions(id)
val part = partitions(p)
val locs = taskIdToLocations(id)
new ResultTask(stage.id, stage.latestInfo.attemptNumber,
taskBinary, part, locs, id, properties, serializedTaskMetrics,
Option(jobId), Option(sc.applicationId), sc.applicationAttemptId,
stage.rdd.isBarrier())
}
}
} catch {
case NonFatal(e) =>
abortStage(stage, s"Task creation failed: $e
${Utils.exceptionString(e)}", Some(e))
runningStages -= stage
return
}
……
}
核心代码的逻辑在于根据传入的stage进行模式匹配,会根据不同类型的Satge创建的不同的Task,那么首先会计算分区得到分区索引集合,然后使用map方法将根据分区id创建xxxMapTask对象,有几个分区id就创建几个xxxMapTask对象。partitionsToCompute是stage.findMissingPartitions()的返回值,那么查看其源码,stage是一个抽象类的引用,调用的这个方法具体的实现在具体的xxxMapStage类中。分别查看一下在resultstage和中的源码:
ResultStage:
override def findMissingPartitions(): Seq[Int] = {
val job = activeJob.get
(0 until job.numPartitions).filter(id => !job.finished(id))
}
ShuffleMapStage:
override def findMissingPartitions(): Seq[Int] = {
mapOutputTrackerMaster
.findMissingPartitions(shuffleDep.shuffleId)
.getOrElse(0 until numPartitions)
}
所以可以看出,partitionsToCompute就是一个分区索引的集合。ResultStage和ShuffleMapStage的numPartitions的值计算方式一样,都是来自于它们所处阶段的最后一个rdd的分区数量值:
job.numPartitions值:
val numPartitions = finalStage match {
case r: ResultStage => r.partitions.length
case m: ShuffleMapStage => m.rdd.partitions.length
}
numPartitions:
val numPartitions = rdd.partitions.length
所以总结一下,应用程序的总任务数量等于每个阶段的最后一个rdd的分区数量之和。